Walker Blackwell wrote:
Hi,
An overview of what I'm doing, I apologize for the long post.
1. I make Piezography ink (piezography dot com).
2. Traditionally we "linearize" the ink curves directly
in the driver so that Gray Gamma 2.2 (Adobe RGB) input values print L*
linear in the print. This serves to create a uniform standard for output
across inks and printers.
Soft-proofing with "Preserve RGB"
makes a VERY VERY good monitor to print match although it does require
slightly more tweaking of the image.
ICCs can optionally be printed on
top of this system for appearance matching from monitor to print with a
given viewing condition.
3. I am looking to create an ICC only workflow where I can preserve this
linearity in the print without requiring recalibration of the ink
placement.
Instead, this shift will happen in the raster data before it
hits the driver. The reason why I'm doing this is because I'm
in the process of building a dedicated Piezography driver and I would
like to get away from the complicated workflow of calibrating individual
ink channels. It's more elegant to have a simple and standard
icc / media-type workflow.
4. I've had some initial success iteratively tuning a set of
measurement files to trick an ICC into behaving linear but this is
complicated and timely. It won't work for a person out in the
world.
5. I have had success in building "fake" RGB / LAB data
out of any number of grayscale measurements. In other words: a user
measures (say) 60 monochrome steps from dark to light and my software
will interpolate those steps to 256 and also to "fake"
low gamut LAB color data and also create the corresponding RGB target
data as well. This cgats file works to build RGB ICC profiles with
i1Profiler or Argyll that print correctly with the monochrome-only print
system.
6. My trouble is in figuring out how to control this appearance curve
that adds too much contrast to the monochrome output. This contrast is
not just with this system but also apparent in traditional RGB print
profiles I create for normal epson drivers and color ink.
7. I have almost decided simply to add the ability to "tune"
the shadows more open when building the fake RGB/LAB/XYZ measurement
data. This would allow for a half-and-half appearance model. This is a
little kludgy so I'm not sure this is the way to go. A
programatic approach would be the best IMO. If anyone has specific
details (or build commands that would get there) I would be all ears.
A few questions related to the project above:
Ok. So if I export a normal cgats measurement file (RGB #s from 0-255),
can I simply divide the RGB #s by 2.55 to get the 0-100 RGB numbers
required by the .ti3? It seems to work in my current testing I just want
to confirm if that's ok to do.
Also, does Argyll support LAB measurements and not just XYZ?
I can
convert from LAB to XYZ for a given observer angle and illuminant (for
the project that I'm doing which is a little odd) but I was just
wondering if Argyll can do this on it's own from LAB. I could
not find it in documentation.
Lastly, I have found that printing evenly spaced LAB values using
Absolute Colorimetric rendering does not preserve linearity in the
print: it actually increases the contrast curve (appearance correction)
significantly.
I think I am probably building this profile wrong somehow
but my blundering is not following a process yet so it will take some
time. i1Profiler and Argyll look to be nearly identical in their
correction curve with argyll increasing shadow separation slightly over
i1Profiler.